Personalized care for complex lives: initial outcomes of a behaviorally-informed complex care intervention

Abstract

Abstract Background 5% of patients account for the majority of healthcare spend, but standardized interventions for this complex population struggle to generate return on investment. The aim of this study is the development and proof of concept of an adaptive intervention to reduce cost and risk of readmission for medically high-risk individuals with any behavioral health diagnosis. Methods A behaviorally-oriented, personalized care service was delivered using a consultative, team-based approach including a physician, counselor, dietitian and social worker in collaboration with nurse care coordinators. Iterative re-conceptualizations informed tailored treatment approaches to prevent acute decompensation while retraining behaviors that impeded recovery. This service was offered to a small set of members of the employee health plan at University Hospitals Cleveland with an existing behavioral health disorder from November of 2020 to March of 2023. 26 members receiving the service were identified and matched with 26 controls using a risk algorithm. Members and controls were then classified as high utilizers (n = 14) or standard utilizers (n = 38) based on utilization claims data. Results Primary outcomes of this study included medical expenditures (delineated as planned and unplanned spend) and readmission risk scores. Compared to risk-matched controls, both planned and unplanned health care expenditures significantly decreased (p < .05) for 7 high utilizers, and unplanned spend only significantly decreased for 19 standard utilizers (p < .05). Risk scores, which predict future spend, decreased significantly for standard utilizers (p < .05), but not for high utilizers. Discussion The value of a behaviorally-oriented personalized care intervention for medically high-risk patients in a commercial insurance population was demonstrated through decreased spend for high utilizers and decreased risk for standard utilizers. Further expansion, refinement, evaluation and scaling are warranted.